While recent advancements in supervised gait recognition have yielded promising results, these approaches rely heavily on annotated walking data, limiting their generalizability to complex environments. This paper presents a self-supervised gait recognition framework using human poses as input to address this challenge, focusing on high-quality pretrained data and self-supervised learning strategies. We first introduce StreamGait, a large-scale, unlabelled dataset that captures in-the-wild distributions of walking sequences. This dataset is curated from Internet livestreams across diverse geographic and environmental scenarios, reflecting variations in real-world camera angles, weather, and pedestrian behavior. Our framework, MirrorGait, conducts self-supervised learning by integration with 2D-to-3D pose reconstruction to synthesize multi-view perspectives for effective 3D-aware contrastive learning. With specific designs of temporal position embedding and gait partition head on a Transformer backbone, the encoder can readily adapt to the periodic and fine-grained nature of gait. Extensive experiments on three widely used gait datasets, Gait3D, GREW, and OUMVLP-Pose, demonstrate that our method, with minimal fine-tuning on the pretrained model, achieves state-of-the-art performance among pose-based gait recognition approaches. The dataset, code, and models are available at https://github.com/BNU-IVC/StreamGait.
论文检索
输入标题、作者或关键词，从 2,578 篇学术成果中精准定位
Event camera, a novel neuromorphic vision sensor, records data with high temporal resolution and wide dynamic range, offering new possibilities for accurate visual representation in challenging scenarios. However, event data is inherently sparse and noisy, mainly reflecting brightness changes, which complicates effective feature extraction. To address this, we propose a self-supervised pre-training framework to fully reveal latent information in event data, including edge information and texture cues. Our framework consists of three stages: Difference-guided Masked Modeling, inspired by the event physical sampling process, reconstructs temporal intensity difference maps to extract enhanced information from raw event data. Backbone-fixed Feature Transition contrasts event and image features without updating the backbone to preserve representations learned from masked modeling and stabilizing their effect on contrastive learning. Focus-aimed Contrastive Learning updates the entire model to improve semantic discrimination by focusing on high-value regions. Extensive experiments show our framework is robust and consistently outperforms state-of-the-art methods on various downstream tasks, including object recognition, semantic segmentation, and optical flow estimation. The code and dataset are available at https://github.com/BIT-Vision/EventPretrain.
Road surface reconstruction is crucial for autonomous driving, providing accurate and up-to-date road geometry for navigation, safety assessment, and infrastructure maintenance. Camera-based methods have become increasingly cost-effective and scalable for this task. However, achieving high-quality large-scale reconstruction remains challenging due to inconsistent observations of the same road surface points. These inconsistencies arise both within single sessions-caused by factors such as vehicle shadows and exposure shifts-and across multiple sessions, where changes in lighting conditions and viewpoints further exacerbate the problem. To address these challenges, we propose MS-Road, a camera-based approach for large-scale road surface reconstruction with strong geometric consistency. MS-Road leveraging self-supervised learning and multi-view consistent constrain to tackle two key issues: inconsistency in road appearance across different observations and inaccurate road height localization. By enforcing spatiotemporal consistency in both geometric and visual aspects, our method produces more reliable reconstructions within and across sessions. Experiments on two public datasets and a real-world dataset demonstrate that our approach achieves robust and high-fidelity reconstruction under diverse and challenging conditions.
Since a building's floorplans are easily accessible, consistent over time, and inherently robust to changes in visual appearance, self-localization within the floorplan has attracted researchers' interest. However, since floorplans are minimalist representations of a building's structure, modal and geometric differences between visual perceptions and floorplans pose challenges to this task. While existing methods cleverly utilize 2D geometric features and pose filters to achieve promising performance, they fail to address the localization errors caused by frequent visual changes and view occlusions due to variously shaped 3D objects. To tackle these issues, this paper views the 2D Floorplan Localization (FLoc) problem from a higher dimension by injecting 3D geometric priors into the visual FLoc algorithm. For the 3D geometric prior modeling, we first model geometrically aware view invariance using multi-view constraints, i.e., leveraging imaging geometric principles to provide matching constraints between multiple images that see the same points. Then, we further model the view-scene aligned geometric priors, enhancing the cross-modal geometry-color correspondences by associating the scene's surface reconstruction with the RGB frames of the sequence. Both 3D priors are modeled through self-supervised contrastive learning, thus no additional geometric or semantic annotations are required. These 3D priors summarized in extensive realistic scenes bridge the modal gap while improving localization success without increasing the computational burden on the FLoc algorithm. Sufficient comparative studies demonstrate that our method significantly outperforms state-of-the-art methods and substantially boosts the FLoc accuracy.
The significance of informative and robust point representations has been widely acknowledged for 3D scene understanding. Despite existing self-supervised pre-training counterparts demonstrating promising performance, the model collapse and structural information deficiency remain prevalent due to insufficient point discrimination difficulty, yielding unreliable expressions and suboptimal performance. In this paper, we present GaussianCross, a novel cross-modal self-supervised 3D representation learning architecture integrating feed-forward 3D Gaussian Splatting (3DGS) techniques to address current challenges. GaussianCross seamlessly converts scale-inconsistent 3D point clouds into a unified cuboid-normalized Gaussian representation without missing details, enabling stable and generalizable pre-training. Subsequently, a tri-attribute adaptive distillation splatting module is incorporated to construct a 3D feature field, facilitating synergetic feature capturing of appearance, geometry, and semantic cues to maintain cross-modal consistency. To validate GaussianCross, we perform extensive evaluations on various benchmarks, including ScanNet, ScanNet200, and S3DIS. In particular, GaussianCross shows a prominent parameter and data efficiency, achieving superior performance through linear probing (<0.1% parameters) and limited data training (1% of scenes) compared to state-of-the-art methods. Furthermore, GaussianCross demonstrates strong generalization capabilities, improving the full fine-tuning accuracy by 9.3% mIoU and 6.1% AP50 on ScanNet200 semantic and instance segmentation tasks, respectively, supporting the effectiveness of our approach. The code, weights, and visualizations are publicly available at https://rayyoh.github.io/GaussianCross/.
In this paper, we present a novel Self-Supervised Learning (SSL) framework tailored for Multi-View Clustering (MVC), which learns cross-view semantic representations with clear clustering boundaries and derives balanced clustering in an end-to-end manner. Concretely, we propose a generative SSL module that learns high-level semantic representations by recovering randomly masked views from observed views. Then the extracted representations are unified via a sample-level local fusion mechanism and projected into a unit-hypersphere space with evenly distributed cluster prototypes such that the pseudo labels can be directly retrieved using cosine similarity. For each sample, we define highly credible positive pairs of the same cluster and negative pairs of different clusters and design a contrastive SSL module to force the sample to move toward its cluster prototype while farther from the other prototypes in the embedding space. Consequently, the representations exhibit clearer clustering boundaries, and the two SSL modules benefit each other. Finally, we further introduce a clustering regularizer to prevent trivial solutions and derive balanced clustering with theoretical guarantees. Comprehensive evaluations over eight benchmark datasets validate the effectiveness of our proposals against ten state-of-the-art MVC methods.
This paper explores silent speech decoding in active brain-computer interface (BCI) systems, which offer more natural and flexible communication than traditional BCI applications. We collected a new silent speech dataset of over 120 hours of electroencephalogram (EEG) recordings from 12 subjects, capturing 24 commonly used English words for language model pretraining and decoding. Following the recent success of pretraining large models with self-supervised paradigms to enhance EEG classification performance, we propose Large Brain Language Model (LBLM) pretrained to decode silent speech for active BCI. To pretrain LBLM, we propose Future Spectro-Temporal Prediction (FSTP) pretraining paradigm to learn effective representations from unlabeled EEG data. Unlike existing EEG pretraining methods that mainly follow a masked-reconstruction paradigm, our proposed FSTP method employs autoregressive modeling in temporal and frequency domains to capture both temporal and spectral dependencies from EEG signals. After pretraining, we finetune our LBLM on downstream tasks, including word-level and semantic-level classification. Extensive experiments demonstrate significant performance gains of the LBLM over fully-supervised and pretrained baseline models. For instance, in the difficult cross-session setting, our model achieves 47.2% accuracy on semantic-level classification and 42.3% in word-level classification, outperforming baseline methods substantially. Our research advances silent speech decoding in active BCI systems, offering an innovative solution for EEG language model pretraining and a new dataset for fundamental research.
Emotion recognition using electroencephalography (EEG) signals has attracted increasing attention in recent years. However, existing methods often lack generalization in cross-corpus settings, where a model trained on one dataset is directly applied to another without retraining, due to differences in data distribution and recording conditions. To tackle the challenge of cross-corpus EEG-based emotion recognition, we propose a novel framework termed Soft Contrastive Masked Modeling (SCMM). Grounded in the theory of emotional continuity, SCMM integrates soft contrastive learning with a hybrid masking strategy to effectively capture emotion dynamics (refer to short-term continuity). Specifically, in the self-supervised learning stage, we propose a soft weighting mechanism that assigns similarity scores to sample pairs, enabling fine-grained modeling of emotional transitions and capturing the temporal continuity of human emotions. To further enhance representation learning, we design a similarity-aware aggregator that fuses complementary information from semantically related samples based on pairwise similarities, thereby improving feature expressiveness and reconstruction quality. This dual design contributes to a more discriminative and transferable representation, which is crucial for robust cross-corpus generalization. Extensive experiments on the SEED, SEED-IV, and DEAP datasets show that SCMM achieves state-of-the-art (SOTA) performance, outperforming the second-best method by an average accuracy of 4.26% under both same-class and different-class cross-corpus settings. The source code is available at https://github.com/Kyler-RL/SCMM.
Audiovisual emotion recognition (AVER) aims to infer human emotions from nonverbal visual-audio (VA) cues, offering modality-complementary and language-agnostic advantages. However, AVER remains challenging due to the inherent ambiguity of emotional expressions, cross-modal expressive disparities, and the scarcity of reliably annotated data. Recent self-supervised AVER approaches have introduced strong multimodal representations, yet they predominantly rely on modality-specific encoders and coarse content-level alignment, limiting fine-grained emotional semantic modeling. To address these issues, we propose VAEmo, an efficient two-stage framework for emotion-centric joint VA representation learning with external knowledge injection. In Stage~1, a unified and lightweight representation network is pre-trained on large-scale speaker-centric VA corpora via masked reconstruction and contrastive objectives, mitigating the modality gap and learning expressive, complementary representations without emotion labels. In Stage~2, multimodal large language models automatically generate detailed affective descriptions according to our well-designed chain-of-thought prompting for only a small subset of VA samples; these rich textual semantics are then injected by aligning their corresponding embeddings with VA representations through dual-path contrastive learning, further bridging the emotion gap. Extensive experiments on multiple downstream AVER benchmarks show that VAEmo achieves state-of-the-art performance with a compact design, highlighting the benefit of unified cross-modal encoding and emotion-aware semantic guidance for efficient, generalizable VA emotion representations.
Object-Centric Learning (OCL) aggregates image or video feature maps into object-level feature vectors, termed slots. It's self-supervision of reconstructing the input from slots struggles with complex object textures, thus Vision Foundation Model (VFM) representations are used as the aggregation input and reconstruction target. Existing methods leverage VFM representations in diverse ways yet fail to fully exploit their potential. In response, we propose a unified architecture, Vector-Quantized VFMs for OCL (VQ-VFM-OCL, or VVO). The key to our unification is simply shared quantizing VFM representations in OCL aggregation and decoding. Experiments show that across different VFMs, aggregators and decoders, our VVO consistently outperforms baselines in object discovery and recognition, as well as downstream visual prediction and reasoning. We also mathematically analyze why VFM representations facilitate OCL aggregation and why their shared quantization as reconstruction targets strengthens OCL supervision. Our source code and model checkpoints are available on https://github.com/Genera1Z/VQ-VFM-OCL.
Recent advances in multi-view 3D multi-person pose estimation have led to significant progress. However, several critical challenges remain, including the limited extraction and integration of multi-domain information, as well as the high annotation costs associated with 3D data in multi-person scenarios. These issues hinder the broader applicability of current methods in complex computer vision tasks. In this paper, we propose a Dense-Sparse Parallel Networks (DSP) framework that jointly leverages spatial, temporal, and frequency-domain information through an adaptive geo-consistency self-supervised strategy. Specifically, we design a multi-view spatial feature extraction module that captures cross-view spatial distributions from dense multi-view feature maps. In parallel, we employ a local-global temporal attention module and a frequency-aware attention module to extract dynamic temporal patterns and localized frequency-domain features from sparse keypoint data. Furthermore, a multi-domain parallel fusion module is introduced to effectively integrate features across all domains, enabling accurate multi-person 3D pose regression. To enhance self-supervised learning, we employ a dynamic view selector guided by reinforcement learning, which reduces the impact of inaccurate pre-trained 2D poses. Experimental results on three benchmark datasets (i.e., CMU Panoptic, Campus, and Shelf) demonstrate that the proposed DSP framework achieves robust and accurate performance, as evidenced by comparisons with other state-of-the-art methods.
Current medical vision-language pre-training models primarily follow two paradigms: report-supervised cross-modal alignment pre-training and reconstruction-based self-supervised pre-training. The former enhances the discriminative power of representations, while the latter facilitates fine-grained representation learning. However, naively combining these two paradigms inherits their inherent limitations: reconstruction-based methods treat all image patches equally during reconstruction, failing to effectively capture critical pathological details-since disease-related regions typically occupy only a small fraction of the image. Meanwhile, alignment-based methods suffer from suboptimal representations due to the presence of false negatives. To address these challenges, we propose a novel pre-training framework that integrates two key components: Pathology-Aware Reconstruction (PAR) and Discriminative Knowledge-Boosted Alignment (DKBA). Through a cascaded training strategy, our framework effectively combines the strengths of both paradigms while mitigating their inherent limitations. During the reconstruction pre-training stage, PAR incorporates pathology-aware priors to enhance the model's ability to capture fine-grained pathological details. In the alignment pre-training stage, DKBA leverages a medical knowledge graph as external supervision to improve cross-modal clustering alignment, thereby reducing the negative impact of false negatives. Extensive experiments on diverse downstream medical imaging tasks including image classification, object detection, and semantic segmentation, demonstrate the superior generalization capabilities of our method. Our code is publicly available at https://github.com/Felix1118/PADKB.
Despite the great success of Large Vision Language Models (LVLMs), their high computational cost severely limits their broad applications. The computational cost of LVLMs mainly stems from the visual sequence of the input, which consists of hundreds or even thousands of tokens. Although existing methods have made progress by removing redundant tokens, they suffer from severe performance degradation with high pruning rates due to the loss of visual information. In this paper, we propose an Adaptive Content Compensation Method (ACCM), which can effectively mitigate the visual information loss via an image caption. Specifically, ACCM comprises two key components: a lightweight caption model and a selector. Firstly the caption model generates question-related descriptions under the guidance of the user instruction. Then the selector further identifies a contextually appropriate caption from multiple candidates. Leveraging self-supervised learning, our modules could be learned efficiently without any human or automated labeling. We conduct extensive experiments across seven benchmarks and the results show that ACCM significantly outperforms existing methods with lower FLOPs (e.g., surpassing SOTA by 20.6% with 6.5% fewer FLOPs) . https://github.com/ASGO-MM/ACCM.
Learning with Noisy Labels (LNL) reduces reliance on high-quality labeled data but often overlooks open-set noise, where noisy samples belong to unknown classes, unlike closed-set noise within known categories.This paper advances LNL by reformulating the problem to incorporate open-set noise through a complete noise transition matrix, enabling a theoretical comparison of its impact on classification error rates against closed-set noise. Our analysis reveals that open-set noise induces smaller error increases, with distinct effects from 'hard' (semantically similar to inliers) and 'easy' (dissimilar) variants. We evaluate entropy-based detection, finding it effective only for easy open-set noise, and propose solutions leveraging vision-language models and self-supervised learning to address hard noise challenges. For empirical validation, we introduce CIFAR100-O, ImageNet-O, and a WebVision open-set test set, enabling robust benchmarking of LNL methods under open-set noise conditions. Recognizing classification accuracy's limitations in capturing model robustness, we advocate out-of-distribution (OOD) detection as a complementary metric. Our theoretical and empirical results highlight the unique challenges of open-set noise, offering new tools and evaluation frameworks to enhance LNL robustness in real-world scenarios.
Ultra-fine-grained visual classification (ultra-FGVC) targets at classifying sub-grained categories of fine-grained objects. This inevitably requires discriminative representation learning within a limited training set. Exploring intrinsic features from the object itself via contrastive learning has demonstrated great progress towards learning discriminative representation. Yet forcingly dividing highly similar categories at the representation level may over-guide the learned feature space, leading to overfitting in the ultra-FGVC tasks. To this end, this paper introduces CLA-Net, a novel contrastive Lie algebra learning framework to address this fundamental problem in ultra-FGVC. The core design is a self-supervised module that performs self-shuffling and masking and then distinguishes these altered images from other images at a second-order representation level. This drives the model to learn an optimized feature space that has a large inter-class distance while remaining tolerant to intra-class variations. By incorporating this self-supervised module, the network acquires more knowledge from the intrinsic structure of the input data, which improves the generalization ability without requiring extra manual annotations. CLA-Net demonstrates strong performance on eight publicly available datasets, demonstrating its effectiveness in the ultra-FGVC task. The code is available at: https://github.com/zichengpan/CLA-NET.
Vision-grounded medical report generation aims to produce clinically accurate descriptions of medical images, anchored in explicit visual evidence to improve interpretability and facilitate integration into clinical workflows. However, existing methods often rely on separately trained detection modules that require extensive expert annotations, introducing high labeling costs and limiting generalizability due to pathology distribution bias across datasets. To address these challenges, we propose Self-Supervised Anatomical Consistency Learning (SS-ACL)-a novel and annotation-free framework that aligns generated reports with corresponding anatomical regions using simple textual prompts. SS-ACL constructs a hierarchical anatomical graph inspired by the invariant top-down inclusion structure of human anatomy, organizing entities by spatial location. It recursively reconstructs fine-grained anatomical regions to enforce intra-sample spatial alignment, inherently guiding attention maps toward visually relevant areas prompted by text. To further enhance inter-sample semantic alignment for abnormality recognition, SS-ACL introduces a region-level contrastive learning based on anatomical consistency. These aligned embeddings serve as priors for report generation, enabling attention maps to provide interpretable visual evidence. Extensive experiments demonstrate that SS-ACL, without relying on expert annotations, (i) generates accurate and visually grounded reports-outperforming state-of-the-art methods by 10% in lexical accuracy and 25% in clinical efficacy, and (ii) achieves competitive performance on various downstream visual tasks, surpassing current leading visual foundation models by 8% in zero-shot visual grounding. Our code is available at https://github.com/kaelsunkiller/ssacl.
Audio-visual speech recognition (AVSR) leverages complementary visual cues to improve speech recognition. However, in real-world scenarios, both modalities may suffer from noise or occlusion. In such scenarios, most existing fusion strategies overlook the variation in modality-specific quality under different degradation conditions. This limitation may lead to dominance of corrupted modality in the fusion process, resulting in worse AVSR performance than unimodal systems, termed as Corrupted Modality Bias (CMB) in this work. To address this, a self-supervised speech representation learning framework, called AV-RISE, is proposed to employ teacher-student self-distillation to robustly reconstruct clean speech representations from corrupted audio-visual inputs. A hierarchical fusion mechanism is designed to progressively refine audio and visual representations by integrating the Suppression and Enhancement Interaction (SEI) module into each layer of the pre-trained encoder. In the SEI module, cross-modal suppression and modality-oriented enhancement are performed to mitigate noise-induced feature inconsistencies, which strengthens the modeling of complementary semantic representations. Extensive experiments on the LRS2 and LRS3 datasets demonstrate that AV-RISE outperforms SOTA AVSR models, especially under extreme degradation. Most importantly, the hierarchical SEI-based fusion effectively enhances reliable semantic representations to mitigate CMB, by evaluating feature similarities between clean and noise samples.
Multimodal graphs, which integrate unstructured heterogeneous data with structured interconnections, offer substantial real-world utility but remain insufficiently explored in unsupervised learning. In this work, we initiate the study of multimodal graph clustering, aiming to bridge this critical gap. Through empirical analysis, we observe that real-world multimodal graphs often exhibit hybrid neighborhood patterns, combining both homophilic and heterophilic relationships. To address this challenge, we propose a novel framework---Disentangled Multimodal Graph Clustering (DMGC) ---which decomposes the original hybrid graph into two complementary views: (1) a homophily-enhanced graph that captures cross-modal class consistency, and (2) heterophily-aware graphs that preserve modality-specific inter-class distinctions. We introduce a Multimodal Dual-frequency Fusion mechanism that jointly filters these disentangled graphs through a dual-pass strategy, enabling effective multimodal integration while mitigating category confusion. Our self-supervised alignment objectives further guide the learning process without requiring labels. Extensive experiments on both multimodal and multi-relational graph datasets demonstrate that DMGC achieves state-of-the-art performance, highlighting its effectiveness and generalizability across diverse settings. Our code is available at https://github.com/Uncnbb/DMGC.
Self-supervised 3D hand pose estimation methods can leverage labeled synthetic data along with unlabeled real-world data for model training, thereby alleviating the reliance on large-scale annotated datasets. Multi-view information fusion is a key factor in the success of these methods. Rule-based fixed fusion methods are simple, efficient, and generalizable, but they neglect the rich visual information in each view. Neural network-based learnable fusion methods can effectively model both intra- and inter-view semantic context, but they tend to overfit to the domain-specific feature of synthetic data and susceptible to interference of domain gaps. In this paper, we decompose multi-view fusion into two components: a learnable confidence estimation stage and a fixed confidence fusion stage. This design not only enables effective use of multi-view semantic cues but also ensures strong cross-domain generalization. To achieve accurate and robust confidence estimation, our method jointly exploits both multi-view pose consistency and pose-to-data consistency. Experiments on three public datasets demonstrate that our approach significantly outperforms existing state-of-the-art self-supervised 3D hand pose estimation methods.
Handwritten Mathematical Expression Recognition (HMER) remains a challenging task due to the structural complexity of mathematical notation and the ambiguity of handwritten symbols-e.g., ''ρ'' vs. ''p'' or ''B'' vs. ''β''. While stroke-based models offer disambiguation via temporal cues, most existing methods are constrained by coarse modality fusion and a lack of fine-grained cross-modal alignment, further hindered by limited annotated data. We introduce Art for Math (Art4Math), a novel framework that leverages the structural richness of human sketches to enhance HMER through fine-grained, modality-aware learning. Art4Math follows a two-stage training paradigm: Art Grounding (A-Grd) and Math Decoding (M-Dec). In A-Grd, the model is trained to reconstruct masked regions of sketches via joint modeling of visual and stroke-level features, encouraging sensitivity to local structural cues and inter-modality alignment. This Art Grounding cultivates a strong inductive bias for parsing abstract, sparse visual forms. M-Dec then adapts this representation to the HMER domain, enabling more precise symbol disambiguation and structural decoding with limited supervision. Extensive experiments across sketch and handwriting-related tasks, including sketch recognition, retrieval, and HMER, demonstrate that Art4Math significantly outperforms existing self-supervised methods, revealing the overlooked synergy between artistic abstraction and mathematical expression.